U.S. patent number 6,933,888 [Application Number 10/867,056] was granted by the patent office on 2005-08-23 for multi-ship coherent geolocation system.
This patent grant is currently assigned to Bae Systems Information and Electronic Systems Integration Inc., Bae Systems Information and Electronic Systems Integration Inc.. Invention is credited to Henry Adler, Melvin Carroll, Richard Schiffmiller.
United States Patent |
6,933,888 |
Schiffmiller , et
al. |
August 23, 2005 |
Multi-ship coherent geolocation system
Abstract
A system is provided for rapidly ascertaining the position of a
pulse train emitter such as a radar using multiple collectors
without requiring more than one platform to measure the same pulse.
Thus time-of-arrival measurements at a number of collecting
platforms are performed, with the positions of the platforms being
accurately ascertainable using GPS data, and with time
synchronization between the spaced-apart collectors performed by
utilizing atomic clocks. In the multi-ship case, geolocation can be
performed on ten milliseconds of data as opposed to 30 seconds of
data for measurements involving a single platform. The subject
system is preferable to conventional time-difference-of-arrival
geolocation systems because those systems require that each of the
collecting platforms measure the same pulse from the emitter, which
severely constrains the flight paths of the collectors, limits the
amount of usable data, and increases the system's sensitivity
requirements.
Inventors: |
Schiffmiller; Richard (Teaneck,
NJ), Adler; Henry (New York, NY), Carroll; Melvin
(Flushing, NY) |
Assignee: |
Bae Systems Information and
Electronic Systems Integration Inc. (Nashua, NH)
|
Family
ID: |
34839070 |
Appl.
No.: |
10/867,056 |
Filed: |
June 14, 2004 |
Current U.S.
Class: |
342/387;
342/442 |
Current CPC
Class: |
G01S
5/14 (20130101); G01S 7/021 (20130101); G01S
19/42 (20130101) |
Current International
Class: |
G01S
1/24 (20060101); G01S 1/00 (20060101); G01S
001/24 () |
Field of
Search: |
;342/387,442,443,444,457,463 |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Phan; Dao
Attorney, Agent or Firm: Tendler; Robert K Long; Daniel
J.
Claims
What is claimed is:
1. A method for improving the accuracy of a coherent
time-of-arrival geolocation system in ascertaining the location of
a pulse train emitter, comprising the steps of: providing multiple
spaced-apart collectors at known locations so as to establish a
long measurement baseline; at each collector, measuring times of
arrival of pulses emitted by the pulse train emitter using
non-snippet based coherent time-of-arrival processing over tens of
seconds to correlate received pulses at different collectors with
associated emitted pulses such that only one received pulse need be
correlated with the corresponding emitted pulse to obtain
geolocation information based on time of arrival of other pulses at
other collectors, thereby to remove any requirement of simultaneous
collection of the same pulse at multiple collection platforms; and,
from an ensemble of measured times of arrival deriving the location
of the emitter.
2. The method of claim 1, and further including the steps of:
locating the collectors aboard separate aircraft; flying the
aircraft along widely spaced-apart flight paths; determining the
location of the collectors along respective flight paths;
establishing a common time reference for all aircraft; and, wherein
the location-deriving step includes using coherently measured times
of arrival of the pulses at the aircraft collectors, whereby the
collectors are instantaneously at different locations relative to
the emitter, such that the time to derive a geolocation is
minimized.
3. The method of claim 1, wherein the measuring step includes
measuring the times of arrival of pulses from the emitter at all
collectors synchronously, thus to minimize processing to achieve a
geolocation of the emitter.
4. The method of claim 3, wherein the measurements of times of
arrival occur over a dwell and wherein all collectors need not
collect pulses over the same dwell.
5. The method of claim 4, wherein the dwell is between 10 and 20
milliseconds and wherein the step of deriving the location of the
emitter is in the millisecond range due to the collection of
time-of-arrival data at the multiple collectors.
6. The method of claim 3, wherein the measuring of the times of
arrival of pulses from the emitter at the spaced-apart collectors
includes effectively using the same clock so as to establish a
synchronous measurement at the widely spaced-apart collector
locations.
7. The method of claim 1, wherein the driving step includes
postulating a launching point corresponding to a rough estimate of
the location of the pulse train emitter; calculating from the
launching point expected times of arrival at the known location of
the collectors; comparing the measured times of arrival of pulses
at a collector with expected times of arrival to generate an error;
in an iterative process, changing the launching point to minimize
the error; and, declaring the launching point to be the location of
the emitter when the error is below a predetermined threshold.
8. The method of claim 7, wherein the postulating step includes
forming a grid swath having grid boxes and locating an initial
launching point in a grid box.
9. The method of claim 7, wherein a point is declared the location
of the emitter only if all of the launching points in the iterative
process have been analyzed.
10. The method of claim 9, wherein the analysis of the launching
points includes establishing which of the launching points results
in the least error, thus to minimize positional ambiguity for the
emitter and selecting from the analysis the launching point having
the least error.
11. The method of claim 9, wherein the analysis of the launching
points includes computing the undoppler-shifted frequency of each
of the measured pulses and declaring a launching point as
establishing the location of the emitter only for launching points
associated with measured pulses which have a frequency matching
that of the emitter.
12. The method of claim 8, wherein for launching points within a
grid box there is a maximum allowed number of iterations for
establishing from the launching points in the grid box an error
below a predetermined error, and further including the step of
locating a launching point in a different grid box if the maximum
number of iterations for the grid box has been exceeded, rather
than selecting a new launch point in the original grid box.
13. A system for improving the accuracy of a coherent
time-of-arrival geolocation system in ascertaining the location of
a pulse train emitter, comprising: multiple spaced-apart collectors
at known locations so as to establish a long measurement baseline;
a receiver at each of said collectors for measuring times of
arrival of pulses emitted by the pulse train emitter using coherent
time-of-arrival processing; and, a processor using coherent
processing in which the same emitted pulse need not be
simultaneously received at said spaced-apart collectors for
deriving the location of the emitter from an ensemble of measured
times of arrival.
14. The system of claim 13, wherein said collectors are located
aboard separate aircraft, said aircraft flying along widely
spaced-apart flight paths, and further including a
position-determining unit for determining the location of the
collectors along respective flight paths; and, a common time
reference for all aircraft, said processor including an algorithm
for coherently measuring times of arrival of the pulses at the
aircraft collectors, whereby the collectors are instantaneously at
different locations relative to the emitter, such that the time to
derive a geolocation is minimized.
15. The system of claim 13, wherein the times of arrival of pulses
from the emitter are measured at all collectors synchronously, thus
to minimize processing to achieve a geolocation of the emitter.
16. The system of claim 15, wherein the measurements of times of
arrival occur over a dwell and wherein all collectors need not
collect pulses over the same dwell.
17. The system of claim 16, wherein the dwell is between 10 and 20
milliseconds and wherein deriving the location of the emitter is in
the millisecond range due to the collection of time-of-arrival data
at the multiple collectors.
18. The system of claim 15, wherein the measuring of the times of
arrival of pulses from the emitter at the spaced-apart collectors
includes a synchronized clock at all collectors so as to establish
a synchronous measurement at the widely spaced-apart collector
locations.
19. The system of claim 13, wherein the processor includes an
algorithm for postulating a launching point corresponding to a
rough estimate of the location of the pulse train emitter, for
calculating from the launching point expected times of arrival at
the known locations of the collectors, for comparing the measured
times of arrival of pulses at a collector with expected times of
arrival to generate an error, for changing the launching point to
minimize the error in an iterative process and for declaring the
launching point to be the location of the emitter when the error is
below a predetermined threshold.
20. The system of claim 19, wherein a point is declared the
location of the emitter only if all of the launching points in the
iterative process have been analyzed.
21. The system of claim 20, wherein the analysis of the launching
points includes an algorithm establishing which of the launching
points results in the least error, thus to minimize positional
ambiguity for the emitter, said algorithm selecting from the
analysis the launching point having the least error.
22. The system of claim 20, wherein the analysis of the launching
points includes an algorithm for measuring the undoppler-shifted
frequency of each of the measured pulses and declaring a launching
point as establishing the location of the emitter only for
launching points associated with measured pulses that have a
frequency matching that of the emitter.
23. The system of claim 19, wherein the algorithm for postulating a
launching point includes a process for forming a grid swath having
grid boxes and wherein the processor algorithm includes a
subroutine for launching points within a grid box, for establishing
a maximum allowed number of iterations for establishing from the
launching points in the grid box an error below a predetermined
error, and for locating a launching point in a different grid box
if the maximum number of iterations for the grid box has been
exceeded, rather than selecting a new launch point in the original
grid box.
Description
FIELD OF THE INVENTION
This invention relates to geolocation and more particularly to a
system for rapidly determining the location of a pulsed waveform
emitter such as a radar using two or more collection platforms.
BACKGROUND OF THE INVENTION
As described in a pending U.S. patent application entitled COHERENT
GEOLOCATION SYSTEM by Richard Schiffmiller, Henry Adler and Melvin
Carroll filed on even date herewith, assigned to the assignee
hereof and incorporated herein by reference, a coherent
time-of-arrival (TOA) geolocation system is described. In one
embodiment a collector aboard an aircraft or other vehicle is moved
from one location to the next to collect data on which a TOA
geolocation of a pulsed emitter such as a radar is made. However,
to establish a long enough baseline for adequate location accuracy,
it takes a considerable amount of time to fly the required baseline
distance, sometimes 30-90 seconds. It would thus be desirable to be
able to cut down the processing time for achieving geolocation.
By way of background, it is tactically important for a military
aircraft that is overflying an enemy territory and detecting pulsed
radiation from a radar to be able to locate the position of the
radar so that the radar can either be destroyed, avoided or
countermeasured. Two classes of time-based methods have been used
in the past to geolocate a radar. The first utilizes time
difference of arrival (TDOA) of radar pulses, measured either
across two antennas of a single aircraft, or across multiple
aircraft. The second measures the time of arrival (TOA) of a
radar's pulses at a single platform in a non-coherent fashion by
averaging data taken from a number of snippets called dwells. This
system exploits the varying inter-pulse intervals due to movement
of the platform from one position to another.
In both of the above cases the accuracy of the geolocation depends
on the baseline between the collectors used to detect the emitted
pulses or the distance the single collector moves during the
geolocation measurement period. Note that the longer the baseline,
the better will be the location accuracy. Prior multi-ship Time
Difference Of Arrival systems, while useful, require that the same
pulses be detected by more than one collector and that the
collectors know which pulses on one platform correspond to which on
the others. The latter requirement, if not met, can lead to
ambiguous geolocations of the radar. The former requirement is even
more severe. If the collectors do not detect the same pulses, the
position of the emitting device cannot be accurately ascertained by
the here-to-fore used methods.
Measuring the same pulses on multiple platforms is difficult to
achieve. There may be physical obstructions that block a platform's
line of sight to a radar so that pulses detected by one aircraft
may not be detected by the others. Also, the collectors' receivers
may not be tuned to the same frequency bands at the same time, and
so will not detect the same pulses from the emitter. Finally, the
collectors may not have the sensitivity to detect a scanning radar
beam in its side lobes or back lobes, and the main beam of the
radar may be illuminating only one collector at a time. As to TOA
systems, prior time-of arrival systems that use non-coherent
processing operate on snippets or dwells of data, with many
snippets of data collected over many tens of seconds of flight in
an attempt to establish a long baseline. This prior method measures
the times of arrival associated with each snippet independent of
the others and then averages the time-of-arrival results. This
approach is called "non-coherent" processing as it does not exploit
any possible long time uniformity or coherency across the snippets
of data. Pulse data is coherent over a period only if there is some
constancy in the radar emission process over that period, e.g., the
pulse repetition interval (PRI) does not change.
As to PRI, typically the emitter's PRI is often purposely varied
depending on the mode of operation or is inherently unstable over
time. Thus, the reason for using short snippets of data in the past
was to assure that the radar's PRI did not change over the
measurement or that there are no gaps in receipt of the pulses,
thereby assuring coherency at least over the snippet.
Note that when only a small snippet of data is considered the
baseline associated with the data is exceedingly short. This means
that any geolocation using the snippet alone will be unacceptably
error-prone. Averaging the times of arrival in a dwell and using
those average values to extend the baseline (non-coherent
processing) typically requires more than 60 seconds of data to
converge and does not produce geolocation results that are nearly
as accurate as coherent processing.
To summarize, processing TOA data non-coherently involves averaging
the time-of arrival results over each short snippet of data and
finding a location for those values. Any coherent processing that
is done occurs only over the short snippets involving short
collection periods or dwells. Because of this, the resultant
geolocation has limited accuracy despite averaging.
SUMMARY OF THE INVENTION
There is, however, a method for speeding up data collection and
markedly increasing accuracy by adapting coherent single-ship TOA
techniques to a multi-ship environment. The subject technique
permits taking data simultaneously at multiple spaced-apart
locations both to increase the baseline and to dramatically
decrease data collection times from tens of seconds to ten or
twenty milliseconds. This is because one does not have to wait for
a single ship to fly from one location to another to establish a
long baseline. The subject technique is also a dramatic advance
over current TDOA techniques in that it does not require that the
collectors measure the same pulses and still achieves the same
geolocation accuracies as the TDOA multi-ship geolocation methods
that do require the measurement of the same pulses on all
platforms. The problems that the subject technique addresses are
the length of time it takes for a collector to move from one
position to another and that the collectors at two different
positions are not receiving the same pulse from the emitter.
The accuracy of the geolocation one can obtain when using multiple
collectors depends on the geometry, with the geometry being defined
as where the radar emitter is with respect to the collectors and
the relative separation of the collectors; the time of arrival
measurement accuracy; and the accuracy of the measurement of the
position of the collecting antenna on each of the platforms. In
general, the geolocation accuracy is better the farther apart the
collectors are from each other and the closer the collectors are to
the emitting radar.
Rather than depending on detecting the same pulse at a number of
different collectors, in one aspect of the subject invention there
are multiple ships, platforms or collectors involved which do not
have to measure the same pulses. Note, the multi-ship algorithm for
use with multiple collectors permits ascertaining the position of
the emitting radar within milliseconds as opposed to tens of
seconds and with geolocation errors measured in tens of meters or
less.
A feature of this invention is the concept that coherently
determining geolocation can be extended to multiple collecting
platforms. As a result, a precise answer can be achieved with only
a single dwell of data (i.e., 10-20 milliseconds) collected at
multiple platforms. This technique yields geolocation errors of
tens of meters or less with milliseconds of data.
It will be appreciated that in order to relate the time-of-arrival
measurement of a pulse to the location of the radar, one must know
the time that pulse was emitted from the radar, the location of the
collector when it received the pulse, and the precise time of
transit (which may be converted to a distance by multiplying by the
speed of light) of the pulse from the radar to the collector. It
will be shown that it is not necessary to measure the
time-of-arrival of any one pulse at more than one of the spatially
separated collectors. The pulse data collected at each platform is
combined and treated as if a single detector measured all the data,
but at different positions. The location of the collector measuring
each pulse is known, so one may consider the situation as if a
single collector was moving to the actual positions of each
collector to receive each pulse that is measured. This requires
that the timing circuits in the pulse collection systems on each
platform are synchronized to one another by the GPS one
pulse-per-second strobe or some other standard, so that they can be
considered as if they are all operating with the same clock.
For example, if there are three collecting platforms that are
spaced twenty miles apart from each other, and each receives a
pulse 10 milliseconds apart from the other, then upon combining the
data with the location of the collector that received it, one can
conceptually think of the situation as if one ship measured the
first pulse, moved 20 miles in 10 milliseconds and collected the
second pulse, and then moved 20 miles again and measured the third
pulse. This "virtual" motion achieves an exceptionally long baseleg
in a short amount of time and can achieve in milliseconds the
performance of a single platform flying for many minutes.
For this "virtual" motion concept to work, it is necessary to know
what pulse each platform is detecting in the sequence emitted by
the radar. In one embodiment, knowing the pulse number, one can
utilize different pulses in the transmission sequence to geolocate
the emitter (i.e., the same pulse does not have to be detected by
more than one collector). The arriving pulse is assigned a number
N, an integer, reflecting the position of the pulse in the emitted
pulse train.
It can be shown that one can ascertain when the first pulse was
transmitted and the underlying pulse repetition interval. In so
doing, one need not provide detection of the same pulse at all
collectors, but need only correlate each received pulse with its
corresponding transmitted pulse from the radar.
When geolocating with a single moving collector, the amount of time
necessary to obtain the location of the emitter is typically
several tens of seconds. Moreover, with a single moving collector
the accuracy of the emitter location measurement is much less than
when one geolocates with several spaced-apart collectors. For
instance, one usually describes the accuracy of a single-ship
geolocation measurement in terms of a percentage of the range, with
a 5% of range accuracy being quite good. This means that at a range
of 20 miles from the emitter, the accuracy is plus or minus one
mile.
However, it will be seen that when utilizing multiple collectors,
the geolocation accuracy error can be reduced to a few tens of
meters or less. This is accomplished when the collectors are spaced
miles apart, creating a significant baseline. Alternatively, the
collectors can be closer to each other, but surrounding the
emitter, and the precision of the geolocation will be on the order
of meters. Each of the collectors must accurately determine its
location when it receives each pulse and can do so by using a GPS
receiver. Further, the geolocation system must provide very
accurate synchronization between each of the collectors. This is
accomplished in one embodiment through the utilization of a local
atomic clock on each platform that is locked to the GPS one
pulse-per-second output such that each of the collectors can be
thought of as having the same clock. Thus a collector can measure
the time of arrival of a pulse from the emitter on the same
timeline as the other collectors.
As a result, for the multi-ship case, all of the collectors are
measuring time in exactly the same way. In short, one can consider
that each of the collectors has the same clock or at least knows
what the relative offset is, which can then be accommodated.
Thus, the different collectors can be thought of as one collector
moving extremely fast from one platform location to another and
measuring the times of arrival of the emitted pulses.
The TOA Process
It will be appreciated that if one can ascertain the time that a
pulse was in fact emitted from the emitter, and in fact if one
measures the time of arrival of that pulse at the collector, then
one can ascertain the time of transit of that pulse. Then, after
converting the time of transit to a distance by multiplying by the
speed of light and knowing the position of the collector when the
pulse arrived, one may deduce the straight line radial distance to
the emitter. By taking multiple time-of-arrival readings for
different collector positions to create a "baseleg," one can
determine the exact location of the emitter. The length of the
baseleg depends on the geometry and the size of the measurement
errors of time and collector position. For the subject invention,
the number of platforms, their relative positions with respect to
the emitter, and the measurement accuracies will determine the
degree of accuracy of the geolocation result.
For multiple pulses, one must know which received pulses at the
collectors correspond to which transmitted pulses at the radar.
Processing the aggregate of data in this way is known as coherent
processing. Coherent processing includes techniques to be able to
specify the underlying constant repetition interval (Q) of the
emitter. The existence of such a parameter Q implies that the data
is time coherent. A constant pulse repetition interval (PRI) also
implies coherence. The coherency is maintained in the processor by
associating an integer N with Q for each pulse being processed.
Thus, for coherent processing, one must be able to know the Q value
of the emitter and use it to know what pulse in a train is being
received. With knowledge of the emitter's Q value and the number of
the pulse, N, received, one can perform geolocation without
requiring the same pulses to be collected by more than one
platform.
For a constant PRI, if the PRI is known, one can determine the
number of any pulse in the sequence from a first pulse starting
point (T.sub.0) by differencing the time-of-arrival of the pulse of
interest and T.sub.0 and dividing by the PRI. For a non-constant
PRI, one must compute the integer, N, for each pulse associated
with Q. By collecting a number of times of arrival of pulses and
comparing them to hypothetical times of arrival generated by
considering hypothetical emitter locations and knowing the sequence
integer, N, for each pulse, one can determine the location of the
emitter. The subject invention finds Q, the integer, N, for each
pulse, the time of transmit of each pulse from the emitter, and the
geolocation of the emitter.
In one embodiment, the system begins by hypothesizing a trial
position for the emitter and a time that the first pulse in the set
was emitted. From that and the known positions of the collectors
when they received each pulse in the set, the algorithm used in the
subject invention generates expected times of arrival of pulses
from the emitter at each of the collectors, for each known position
of the collectors. If the measured times of arrival are
sufficiently close to the expected times of arrival, then the
position of the emitter is that which was hypothesized.
In practice, one is provided with a rough indication of the
location of the emitter through coarse angle-of-arrival and range
measurements. With a rough guess as to emitter position, one
postulates a grid in the form of a swath and then hypothesizes a
starting location within the grid.
At the outset, the subject geolocation algorithm establishes a
first launching point in the grid, calculates the expected times of
arrival of pulses at the various known collector locations and then
compares the actual measured times of arrival with those that are
expected. If the error between the two is not less than a threshold
value, the hypothesized position is driven by a gradient descent
method (Newton-Raphson) to a potentially more accurate location for
the emitter. If, at any point during a fixed number of iterations
of the gradient descent algorithm, the error drops below a
predetermined threshold, the hypothesized location of the emitter
is the declared location of the emitter. If after the fixed number
of iterations, the error is not below the threshold, then a new
grid box position is used to launch the algorithm and the process
is repeated.
Ambiguities
Once a location is found that produces TOA errors less than the
threshold value, that location is stored and a new grid box
position is used to re-launch the algorithm and the process is
repeated. Every valid location derived from the grid launch points
represents an ambiguous location of the emitter. Ambiguities occur
because of the large spacing of the collectors, and can occur even
when there are no measurement errors at all.
In one embodiment, ambiguities are effectively eliminated by
exploiting the velocity of the collectors, even though they hardly
move at all during the 10 or 20 milliseconds of the data collect.
The frequency of the pulses is Doppler-shifted due to the
radial-component of the velocity vector of the collector with
respect to the emitter. If the radial component of the aircraft's
velocity vector is towards the emitter, the measured frequency is
higher than it is at the radar; and if it is away from the emitter,
the measured frequency is lower. Each ambiguous location is used to
compute the Doppler shift for each measured pulse for each
collector. The shift is then removed from each measured frequency.
For the correct geolocation solution, all the pulse frequencies
should then result in the same value, namely the frequency of
transmission at the radar. The incorrect solutions will have a
potpourri of resultant frequency values for the pulses and are
ignored.
It can be shown that for a multi-ship collector system in which the
measured pulse data of each of the multiple-ship collectors is
combined with the collector location and treated as if one ship
performed all the collections, then the subject system can
ascertain the emitter location within meters as opposed to miles.
It can further be shown that one only needs short bursts of pulses,
for instance ten milliseconds, in order to accomplish the above
measurement. While establishing an accurate time base for the
flying platforms as well as establishing the precise location of
the platforms at any given time is a non-trivial matter, it is
possible with atomic clocks and GPS coordinates to ascertain the
exact positions of the collectors and to be able to have the
collectors function as if they were all functioning from the same
clock.
In one embodiment, during the data collection phase, every received
pulse is tagged with a time of arrival. Note that one does not have
to have the same sensitivity for the receivers in the subject
invention as in conventional time-difference-of-arrival systems
because the collectors do not have to detect the same pulses. If a
main lobe of the radar beam is pointing at one platform and the
sidelobe of the radar beam (much weaker in power than the main
lobe) is pointing at a second platform at one instant, then the
conventional system would require a higher degree of sensitivity so
that the platform can see the pulse in the sidelobe that the other
platform measures in the main lobe. Because this is not necessary
in the subject invention, the subject invention allows about 5 dB
less sensitivity for the receiver systems on the platforms than the
conventional TDOA system.
In summary, a system is provided for rapidly ascertaining the
position of a pulse train emitter such as a radar using multiple
collectors without requiring more than one platform to measure the
same pulse. Thus time-of-arrival measurements at a number of
collecting platforms are performed, with the positions of the
platforms being accurately ascertainable using GPS data, and with
time synchronization between the spaced-apart collectors performed
by utilizing atomic clocks. In the multi-ship case, geolocation can
be performed on ten milliseconds of data as opposed to 30 seconds
of data for measurements involving a single platform. The subject
system is preferable to conventional time-difference-of-arrival
geolocation systems because those systems require that each of the
collecting platforms measure the same pulse from the emitter, which
severely constrains the flight paths of the collectors, limits the
amount of usable data, and increases the system's sensitivity
requirements.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other features of the subject invention will be better
understood in connection with a Detailed Description, in
conjunction with the Drawings, of which:
FIG. 1 is a diagrammatic illustration of a single collector system,
shown collecting data at two points along the flight path of an
aircraft, with the pattern of changes of inter-pulse intervals
resulting in the geolocation of the pulse-emitting source;
FIG. 2 is a diagrammatic illustration of a two-ship embodiment of
the subject invention, illustrating that the collection of data on
a single moving platform can be achieved by two platforms at
separate locations over such a short time that the platforms have
not moved any appreciable distance;
FIG. 3 is a diagrammatic illustration of an embodiment of the
multi-ship collection system of the subject invention in which each
ship receives inputs from GPS satellites that it uses to locate
itself and to synchronize its onboard clock with the other ships,
each ship measuring the time-of-arrival of pulses from a radar and
outputting those times and its location at the time it measured
each pulse to a geolocation computer for the computation of the
location of the radar;
FIG. 4 is a diagrammatic illustration of the pulse repetition
interval between adjacent pulses in a pulse string or sequence
generated at the pulsed transmitter;
FIG. 5 is a diagrammatic illustration of the measured time of
arrival of received pulses for three different collector standoff
distances, showing different transit times between the transmitted
and received pulses;
FIG. 6 is a formula that is used in geolocating a pulsed
emitter;
FIG. 7 is a diagrammatic illustration of the geolocation of an
emitter in terms of geometry involving vectors .DELTA.R.sub.1,
.DELTA.R.sub.2 and .DELTA.R.sub.3, with the vectors referenced to a
point at the center of the earth;
FIG. 8 is a flow chart illustrating the subject processing method
for geolocating a pulsed emitter that utilizes a hypothesized
position of the emitter as a starting or launching point, the
generating of expected times of arrival of pulses based on the
hypothesized emitter position given known collector positions, the
measurement of actual times of arrival of pulses at the collectors,
the comparison of the measured with the hypothesized times of
arrival and the use of a gradient descent method to modify the
hypothesis to minimize the error between the expected times of
arrival and the measured times of arrival; and,
FIG. 9 is a diagrammatic illustration of a grid swath generated
from rough emitter position data, in which a launching point is
located within the grid, with the launching point being modified in
accordance with a gradient descent method to change the launching
point to one which minimizes the error.
DETAILED DESCRIPTION
Referring now to FIG. 1, in a single-collector embodiment of the
subject invention, an aircraft 10 at a position x.sub.1, y.sub.1,
z.sub.1 at a time t.sub.1 flies along a path 12 so that, as
illustrated at 10', it occupies a position x.sub.2, y.sub.2,
z.sub.2 at a time t.sub.2. It should be noted that the figure shows
the arrival of two pulses at the collector. In practice, many more
pulses must be collected such that the time to move from the
initial location to the final location can be in excess of 30
seconds to establish an adequate baseline between the first pulse
collected and the last pulse collected.
An emitter 14 produces a string or sequence of pulses 16 at a pulse
repetition interval or PRI of between 205 Hz to 300 KHz for typical
radars. It is noted that the aircraft when in position 10 is at a
distance d.sub.1 from emitter 14, whereas when the aircraft is at
position 10' it is at a distance d.sub.2 from the emitter. Note
also that the aircraft at position 10 receives a pulse 18
designated pulse.sub.1, whereas at time t.sub.2 the collector at
position 10' receives a pulse 20 designated pulse.sub.2.
Key to this invention is that the notion of one aircraft moving
over time to a second position to collect additional pulses is
equivalent to having two aircraft instantaneously at the two
positions collecting pulses. This is the basis of the multi-ship
embodiment of the subject invention.
For the aircraft at position 10, in the graph below, the pulses at
the emitter are illustrated at 22, whereas the first pulse in the
sequence arrives at the collector at a time of arrival designated
by dotted line 24. The time interval between the transmission and
receipt of the first pulse is illustrated by double-ended arrows 26
and corresponds to the transit time.
As can be seen by dotted line 28, when the collector is closer to
the emitter, received pulse 28 will arrive in a shorter amount of
time than when the collector is a distance d.sub.1 from the
emitter. This means that as the collector moves closer to the
emitter, the inter-pulse interval between received pulses will be
shorter than the interval between those pulses at the
transmitter.
As illustrated at 30, knowing a pattern of changes in the
inter-pulse intervals, one can establish the geolocation of the
emitter as illustrated at 32.
Here it will be seen that in the two-collector interpretation of
FIG. 1, the collectors may or may not see the same pulses at their
respective locations. As will be described, it is possible to
ascertain which pulses in pulse train 16 pulse 18 and pulse 20 are;
and for this reason it is possible to accumulate an ensemble of
times of arrival from each of the multiple ships and combine them
as if one ship had collected them. Associated therewith will be a
pattern of changes in the inter-pulse intervals for which only one
emitter position on the surface of the earth will exist.
The subject algorithm solves the location not by a direct solution
of an equation but rather by postulating the position of the
emitter and what set of time-of-arrival results is expected at the
various positions of the platforms. One thereafter comes up with an
error that is minimized by driving the hypothesis to another
location.
In operation, assuming that each emitter is fixed at a particular
location, one collector receives the first pulse with a certain
time of arrival based on its distance from the emitter. The second
pulse may then arrive at a collector that is at a difference
distance from the emitter and there is a time interval between the
first two pulses. This is called the inter-pulse interval. The
inter-pulse interval is derived from time-of-arrival measurements.
Typically, there is a certain fixed interval at which the emitter
generates pulses, for instance every 250 microseconds. Since the
pulses are detected in general by collectors at different
positions, the times of arrival are either sooner or later than the
would have been had they all been collected by a single stationary
collector. If the platform measuring the second pulse is closer to
the radar than the platform that measures the first pulse, the
second pulse is going to come in a little bit sooner than the 250
microseconds between the pulses, i.e., the PRI of the radar. Thus,
the system is going to see the second pulse at a time a little bit
less than 250 microseconds from the first pulse. Then if a third
platform that measures the third pulse is farther from the radar
than either of the first two, the system will measure a longer than
250 microsecond interval between the second and third pulses. If
the fourth pulse is measured by the first collector, yet another
inter-pulse interval will be measured, and so on.
What one therefore obtains is a series of inter-pulse intervals
that are changing. One can then generate a pattern of the changes
in inter-pulse intervals. This is done by measuring times of
arrival and by comparing this pattern to a pattern of times of
arrival generated from a hypothesized point. When one compares the
measured times of arrival with the expected times of arrival from
the hypothesized position, one can develop a number representing
error. One can use a gradient descent method to modify the
hypothesized location of the emitter to drive the error to zero.
When the error becomes sufficiently small, the associated
hypothesized location will satisfy all of the times of arrival.
This unique spot is identified as the actual emitter location. If
the gradient descent algorithm does not converge, then a new point
on the grid is selected as a launch point and the process is
repeated. Once a solution is found, the process is repeated at the
next grid point. If after all grid points have been examined, there
are more than one solution, these are ambiguities and an ambiguity
resolving routine is run. Ambiguities occur because of the large
spacing of the collectors, and can occur even when there are no
measurement errors at all.
In one embodiment, ambiguities are effectively eliminated by
exploiting the velocity of the collectors, even though they hardly
move at all during the 10 or 20 milliseconds of the data collect.
The frequency of the pulses is Doppler-shifted due to the
radial-component of the velocity vector of the collector with
respect to the emitter. If the radial component of the aircraft's
velocity vector is towards the emitter, the measured frequency is
higher than it is at the radar, and if it is away from the emitter,
the measured frequency is lower. Each ambiguous location is used to
compute the Doppler shift for each measured pulse for each
collector. The shift is then removed from each measured frequency.
For the correct geolocation solution, all the pulse frequencies
should then result in the same value, namely the frequency of
transmission at the radar. It can be shown that the incorrect
solutions will have a potpourri of resultant frequency values for
the pulses and can be ignored.
In terms of accuracy, the longer the baseline over which the
collectors are positioned, the greater the accuracy of the
geolocation calculation, but the greater the possibility of
ambiguities. By using the ambiguity resolver, one obtains a very
accurate geolocation in a very short amount of time with no
ambiguities. With a single ship system flying a straight-line path,
one has to fly a fair distance, for instance, several miles, along
the baseline to achieve greater accuracy.
Note that if the collector is an antenna on an aircraft, the exact
position of the antenna defines the position at which the data is
collected. The position of the antenna will vary depending on the
orientation of the airplane so that one has to factor in the
position of a GPS point on the airplane, roll, pitch and yaw
information to be able to accurately calculate where the antenna
actually is. To the extent that the positions of a collector are
not instantly known, i.e., the so-called navigation messages for
the collectors come at different times than the times of arrival of
the pulses, one must interpolate the positions of the airplane to
the time of arrival of each pulse.
The formula for obtaining the hypothesized time of arrival of a
pulse is as follows:
Here .DELTA.R contains the x, y, z emitter location information
(unknown) and the location of the aircraft that measured the pulse
at the time of arrival of that pulse (known). As can be seen, one
needs to know the exact pulse-to-pulse interval (PRI) for the
radar. It cannot be assumed, for instance, that the pulse
repetition interval of the radar is constant. In point of fact,
radar pulse repetition intervals are not always constant. For the
subject system to work accurately it is therefore necessary that
the pulse-to-pulse interval be quickly ascertainable. If not, the
processor must run through all the possible PRIs until the error
between the hypothesized and measured times of arrival is below the
appropriate threshold. It can be shown that there is a quantity Q
associated with every radar that provides an estimate of the
instantaneous true PRI of that radar.
It is also important to be able to know in a string or sequence of
pulses which pulse a particular collector is detecting. While it
may be impossible to know which is the first pulse from a radar,
one can ascertain which pulse is first to arrive at one of a number
of collector positions. Then assuming that this is the first pulse,
one needs to be able to ascertain what the number of each
subsequent pulse is--fifth, seventh, 25.sup.th, et cetera. It can
be shown that it is possible to be able to ascertain what the pulse
number is as an integer N related to Q, which represents the pulse
number.
Given the fact that one can ascertain the pulse repetition interval
or a quantity equivalent thereto, one can, through the
above-mentioned iterative technique, drive the algorithm so as to
minimize the error between the computed and measured TOAs, thus to
be able to specify the emitter location when the error is below a
predetermined threshold.
In the iterative solution to the geolocation problem, an emitter
position is first postulated and the actual time-of-arrival
measurements are matched with the expected time of arrivals given
the postulated position. If one knows that a collector is at a
given location and if one postulates the position of the emitter,
one can calculate exactly what the times of arrival of the pulses
are supposed to be. The question then becomes as to how well the
observed set of times of arrival agrees with those that are
generated as a result of the postulated position. An error is
computed between the times of arrival from the hypothesized
position and the actual measured pulses, with a Newton-Raphson
algorithm utilized to correct the hypothesized position to minimize
the error.
It can be shown that in Equation 1 there are 5 unknowns. T.sub.0
(the time the first pulse is transmitted from the emitter) is
unknown, the x, y, z coordinates of the emitter are unknown, and
the PRI is unknown. Another parameter that equates to the true PRI,
namely Q, is unknown and there is an integer N that identifies what
pulse is arriving at a collector. One can know N and can choose a
trial value of Q. This value of Q can be refined in the iterative
procedure.
Q is selected in an averaging process to correspond to a calculated
pulse repetition interval for the pulse train emitted by emitter
14. As noted hereinabove, radar pulse emitters either vary their
PRIs or have a jitter or instability in their pulse trains, so that
the pulse repetition intervals varying over a period of time.
Referring to FIG. 2, two collectors 40 and 42 are shown at two
positions, x.sub.1 y.sub.1 z.sub.1 and x.sub.2 y.sub.2 z.sub.2,
respectively. The first receives a pulse at t.sub.1 and the second
at t.sub.2 from emitter 14. Each collector uses GPS 44 to compute
its position and synchronize its clock 46 so that the system may be
considered as if one platform collected both pulses after moving
from one position to the other.
Referring to FIG. 3, GPS satellites 56 provide platforms 50, 52 and
54 with location and precise time information in terms of clocks
58, 60 and 62, respectively. The pulses measured at each platform
from emitter 14 are sent with the associated platform location to a
geolocation computer 70 that performs the geolocation computation.
All TOAs are combined in time order and a set of integers is found
for the ensemble, together with the parameter Q. A search swath 72
is used to generate coarse hypothesized locations of the emitter.
Using the hypothesized location of the emitter, the position of
each platform when it receives each pulse and the integer N of each
pulse and Q value, the hypothesized time-of-arrival is computed
according to Equation 1. The reference time is the time-of-arrival
of the first pulse. The set of hypothesized times-of-arrival is
compared with the set of measured TOAs to find the geolocation. All
valid results are input to an ambiguity resolver 74, which selects
the correct geolocation of the emitter.
Referring now to FIG. 4, as mentioned hereinabove the pulse
repetition interval of the emitter is critical to the accuracy of
the geolocation process. The pulse repetition interval is defined
as the interval between transmitter pulses, here show at 76.
Referring to FIG. 5, if one can accurately establish the pulse
repetition interval at the emitter, then the time difference
between a transmitted pulse 76 and a received pulse 78 establishes
the distance .DELTA.R.sub.1 /C between the emitter and the
particular collector. An can be seen, for a distance d.sub.2 which
is closer than d.sub.1, .DELTA.R.sub.2 /C as illustrated at 80 is
closer in time to transmitted pulse 76 than is received pulse 78 to
its corresponding transmitted pulse 76.
Likewise, for a collector antenna at a distance d.sub.3 from the
emitter, the received pulse 82 at d.sub.3 is received at some time
later than the time of the emitted pulse 76. The result is that one
can establish .DELTA.R.sub.1 /C, .DELTA.R.sub.2 /C, .DELTA.R.sub.3
/C, etc. by measuring the times of arrival of pulses at multiple
distinct distances from the emitter and comparing with hypothesized
arrival times, as described hereinabove.
How the position of the emitter is obtained can be seen by
considering the equation of FIG. 6 in which the time of arrival of
an ith pulse is computed as a time T.sub.0 +.DELTA.R.sub.i
/C+(n.sub.i -1) PRI. .DELTA.R.sub.i is the difference between one
of the collectors' positions (known) and the emitter's position
(unknown). n.sub.i (the number of the ith pulse from the first one
received in the collect) and the PRI are calculatable. Thus the
equation contains five unknowns: T.sub.0, x, y and z of the
emitter, and the PRI. The PRI is treated as unknown even though an
initial guess is provided as a result of determining n.sub.i. At
least five equations (corresponding to five pulses) would be needed
to solve for the five unknowns. Since the measurements contain
errors, the equations cannot be solved exactly, and the error
minimization technique described above is employed to solve for the
unknowns. The initial guess for the PRI was mentioned, the initial
guess for the emitter position was described above, and the initial
guess for T.sub.0 is simply the time of arrival of the first pulse.
The Newton-Raphson gradient descent algorithm applied to the
differences between the TOAs computed by the equation in FIG. 6 and
the measured TOAs nudges the values of all five unknowns until a
minimum value is found.
Referring to FIG. 7, the geometry involved is shown by referencing
all vectors to a common reference, namely the center of the earth
as illustrated at 84. Note that collectors on three aircraft 86, 88
and 90 are at different distances from emitter 14. This is
described by vectors .DELTA.R.sub.1, .DELTA.R.sub.2 and
.DELTA.R.sub.3. Knowing .DELTA.R.sub.1, .DELTA.R.sub.2,
.DELTA.R.sub.3, etc., one can derive the emitter position x.sub.e,
y.sub.e, z.sub.e.
Referring now to FIG. 8, in one embodiment of the subject invention
one first hypothesizes the position of an emitter as illustrated at
100. This is called the launching point. The hypothesized position
of the emitter is established from a grid 102; which is in turn
derived from an angle-of-arrival estimate 104 and a coarse range
estimate 105.
The hypothesized position of the emitter is utilized at 106 to
generate expected times of arrival of pulses from the hypothesized
emitter position given known collector positions. The result is
that for a given launching point, unit 106 generates an entire set
of hypothesized times of arrival.
As illustrated in 108, one measures actual times of arrival at the
various collector positions and at 110 compares the expected times
of arrival with the actual times of arrival. The difference is an
error which can be reduced as illustrated at 112 utilizing a
gradient descent method, the common Newton-Raphson method, to
modify the hypothesized position of the emitter to minimize the
error between the expected times of arrival and the measured times
of arrival. One utilizes the Newton-Raphson technique to change the
hypothesis as illustrated at 114 so as to change the launching
point to one which will result in expected times of arrival more
closely approximating the measured times of arrival.
The process is iterative, with the launching points being moved in
a direction that minimizes error, up to a maximum number of
iterations. When the error in any iteration is less than a
predetermined threshold as illustrated at 116, the output is the
emitter position, which is the position indicated by the last point
selected by the Newton-Raphson algorithm that resulted in the error
going below the predetermined threshold. In one embodiment, if the
algorithm fails to find a point in a grid box that drives the error
below a threshold value or fails to find a stable error in a grid
box before exceeding a predetermined maximum number of
Newton-Raphson iterations for the grid box as determined by
decision box 118, no geolocation answer is output and a different
launch point in a different grid box is selected.
In one embodiment, because of possible ambiguities, the above
process is repeated until all points in all of the grid boxes have
been used as launch points, as determined by decision box 117.
Having this entire array of launch points, the system analyzes the
results and selects the unambiguous solution amongst the results
from all of these launch points, i.e., all of the launch points in
all of the grid boxes.
The above process runs through all the grid launch points because
in the multi-ship case, in some instances where the collectors are
50-100 miles apart, there may be ambiguities. That is, multiple
grid launch points may yield different geolocation solutions that
are all viable. Special ambiguity processing, shown in 74 of FIG. 3
and described above, is then invoked to select a single location as
the final geolocation solution.
Referring to FIG. 9, in order to set a launching point, one
generates a swath 120 of grid boxes that is based on coarse range
measurements and coarse angles of arrival measurements. How fine
the grid is to be is a function of the expected pulse repetition
interval or, as mentioned before, Q. Using the above-mentioned
gradient descent method the launching point is moved in a grid box,
and the error is assessed. When the error is below a predetermined
threshold, the location of the emitter is known. The above
techniques of FIG. 8 are used to allow the Newton-Raphson algorithm
to converge by properly selecting launching points that are close
enough to the emitter location.
With the multi-ship system, the collectors do not have to move at
all because one achieves differences in distance simply by the fact
that the collectors are in different positions. Thus, the subject
invention may be applied to situations where the collectors are
stationary. This is the case where ground troops pre-position
spaced-apart collectors at known locations. As a result, one need
not use aircraft to locate enemy radars, but rather ground-based
units can be deployed. In the stationary multi-collector case, the
ambiguity resolver 74 in FIG. 3 simply examines the error in the
cost function (the sum of the differences between the hypothetical
and measured TOAs) for each candidate solution and chooses the one
with the smallest error as the location of the radar.
While the present invention has been described in connection with
the preferred embodiments of the various figures, it is to be
understood that other similar embodiments may be used or
modifications or additions may be made to the described embodiment
for performing the same function of the present invention without
deviating therefrom. Therefore, the present invention should not be
limited to any single embodiment, but rather construed in breadth
and scope in accordance with the recitation of the appended
claims.
* * * * *